Hierarchical Bayesian Method in the Study of Individual Level Behavior- In the Context of Discrete Choice Modeling with Revealed and Stated Preference

ISBN-13: 9783836423533 / Angielski / Miękka / 2007 / 80 str.

Xiaojing Dong

Opis

A major breakthrough in travel demand modeling in the early 1970s wasmodeling based on disaggregate (individual) level data (McFadden 2001).Although the disaggregate model focuses on individual level behavior, theestimated model parameters are fixed across individuals. To incorporateunobserved taste variations across individuals, recent developments allow forthe parameters to vary across individuals, such as the Mixed Logit model,where the parameters are assumed to follow a distribution. The mixed logitmodel recognizes the differences among individuals, but it does notdistinguish individuals who respond differently to travel service changes. Thisstudy focuses on the application of the Hierarchical Bayesian method toobtain individual level inferences. We demonstrate the advantage of thismethod by obtaining a more reasonable distribution of value of travel timerelative to the distribution obtained from the mixed logit model. In addition,the HB method helps us to combine information from both revealed andstated preference data, where the revealed preference data is limited toproperties of only the chosen alternatives.